Questions from my son and the future of work

Yesterday, I picked my son up from school.

He usually gets annoyed if I bring up AI or debate, so I try to avoid those topics.  And sometimes there are no topics because he’s in high school and he often doesn’t want to talk.

Yesterday, he brought up AI. He said a lot of students at school were using ChatGPT to do their schoolwork. I said it can be helpful if it is used properly.

I asked him if his teachers talked about it; he said no.

I took a risk and continued the conversation. I told him that if you just use the free version, you don’t see the power of it—not just ChatGPT, but all of AI, and that AI was more than ChatGPT. I told him how there were a lot of recent advances in drug development and materials discovery . I told him about the article that was published about how AI was used to help lead a science experiment and that it acted in at least a semi-autonomous way. He is interested in becoming a doctor, so I hoped this would help him relate.

He asked me if AI was aware of its existence. That was a bit “out of the blue” and indicated to me that he’s been thinking about AI.

I said it was unlikely, but that there was a debate over whether the more advanced AI systems were just repeating back language they had been trained on or if they were reasoning. I said that most people thought they had some low level of reasoning ability and that there was more and more evidence of it, especially in the newly released “Google” model, but that not everyone thought that.

He went back to consciousness without using the word, so we had a brief discussion about how he didn’t know if I was conscious. He laughed, but he got it after I gave a few examples and he gave it some thought.

He asked about AI and jobs. I said I had just written a blog about how a lot of people would lose their jobs, but that people often find new work and that hopefully there would be new jobs eventually, though that was a bit uncertain if the AI could do the new jobs as well. I did say he might just have more time for fun and that there would be arguments in society over how wealth should be redistributed.

He then talked about some things he was thinking about related to college admissions.

But then he sort of asked me the kicker: When I’m your age, what will the world look like? I said no one knows, but that it won’t look like this. This is no different than what Sam Altman said in his Time Magazine interview:

This is just different from anything else. Society is going to fundamentally change.

I very briefly told him about quantum computing and that once we have working quantum computing, how society functions will be completely unknown, and it is impossible to know what work or society will even look like then.

He asked if there would be robots around him all the time. I said yes.

I dropped him and hopped on a call about an AI class I’ll be teaching in the spring.

After I hung up from the call, I thought about two things.

One, many kids wonder about AI (I’ve had some incidental conversations with some others), and very few have anyone who knows anything about it to talk to.

Two, I remembered that I said I’d comment on Microsoft’s “Future of Jobs” report because I think there is some positive potential there. Assuming, of course, that as educators we react to the information. Sometimes I feel like we are the government policy-makers who keep ignoring the reports that say the earth is warming.

I was about to get started, but I got distracted. I saw the new demo of Gemini using its reasoning and math capabilities to help teach math and sent it to my son. If it’s reasoning and not predicting text, it won’t hallucinate.  And it means a free, non-hallucinating math tutor that will give you the answer just went global. There are some countries where Google is not accessible, but the technology will spread to other applications that are accessible in those countries.

I thought I was ready to write about the future of work, but I got distracted again. I saw a new Reid Hoffman podcast drop, and it’s an interesting interview.

I’ll mention that he reinforced what Sam Altman said.

(AI) is somewhere between the largest technological transformation of our lifetimes and the largest technological transformation in the history of humanity.

And that, he discussed how many mid- and upper-level management/knowledge workers, including many accountants, are going to lose their jobs, and those who want to keep their jobs need to learn how to use AI. There is no alternative if you want to remain employed.

Finally, to the future of work. The new Microsoft Future of Work report makes it obvious why these individuals must learn to use AI. It cites the common research that has been shared many times before on this blog and elsewhere. We are generally looking at 30-80% productivity increases when AI is utilized, depending on the profession. Relatedly, they point out that CoPilot in M365 (their product) saves employees time (hence the productivity gain). These gains are strongest when individuals learn how to prompt well, and LLMs have become prompt optimizers.

The biggest boost is for the weaker workers. The report cites multiple studies that show that LLMs benefit the weakest workers the most. For better or worse, it’s an intelligence leveler, and it’s why many, including Sam Altman, say the economic value of intelligence will collapse in a world of AI. Are you smarter than your co-workers? Very soon, that may not matter much at all.

People who use the tools wrongly and rely on the tools when the tools are wrong do worse work, and people who use the tools wrongly and rely on the tools when the tools are wrong do worse work. Overreliance also undermines performance. On the plus side, digital conversation capture technologies can capture the knowledge produced in conversation, and the models can boost human creativity.

So, learn how to use the tools for what they are supposed to be used for and avoid using them for what they shouldn’t be used for.

---

As educators, what does this mean?

First, our students need to learn how to use AI tools and how to use them properly. A recent study showed that will get employees a 16% wage boost, as much as having a PhD. And you don’t even need a college degree to get that boost. Students who cannot use these tools well will struggle to find work, especially over time as knowledge about how to use them well will be assumed. Relatedly, using the tools includes collaborating with them.

Second, meta-cognitive skills are critical for future workers to develop.  Being prepared for the AI world isn’t just about how to use AI, and generative AI in particular; it’s about how to build useful AI models and how to interact with them. Microsoft says these require metacognitive skills such as the ability to argue/understand arguments, analyze, integrate, be self-aware, etc. And not only does AI require that, but it also helps enhance those skills when used properly.

The Microsoft report provides some studies in support of its claims, and we go into depth on this issue in our report on human deep learning. This is something schools need to think more about as they help prepare for the AI world; that preparation can’t just be limited to teachers and students using magic AI tools.  It does argue that AI tools can help support metacognition when used properly.

There is a tremendous opportunity for using AI tools in tutoring and support (see the Google video above for an example).

On a side note, it is awesome that Microsoft referenced Stephen Toulmin and the importance of teaching argumentation and debate. Yes, every school should have a debate team, and all students should learn how to debate.

And the debate itself has to shift a bit.  We spend a good amount of time teaching debaters how to research, but research is everywhere, and, as the report notes, “analyzing and integrating may become more important skills than searching and creating.” Search is still important, but it should consume a small fraction of the time compared to the other tasks.

Third, according to the report, there are gains in multilingual LLM performance, but more is needed. Unaspiringly, they only reference the limited gains made in this by Microsoft and OpenAI. Meta’s gains, which they do not reference, are quite impressive, allowing conversation between two people in different languages while wearing Meta’s glasses. This will force schools to re-evaluate how and why foreign languages are taught.

Fourth, AI tools will change how social science and even science research are done. Schools need to help students understand how to integrate AI into scientific inquiry.

Fifth, the introduction “of AI into any space is an inherently socio-technical process.” I’ve said this many times, but this can’t only be about what edtech specialists think (as we are seeing a lot of in K–12) but about everyone in an educational space.

---

To pull it back to the beginning, as we move into 2024 I hope all educators take seriously the idea of working to prepare students for the world they will live in.

This world will be completely different than the one we live in now.

Think about it. We ended last year with public awareness of an AI that could write essays and had some knowledge. We end this year with an AI that can write and speak in more than 100 different languages, knows a lot more than any human (and is significantly more accurate than it was a year ago), can create and perform music, produce news broadcasts and short films, discover drugs and materials, create immersive environments, tutor students, do higher-level math and statistical analysis, and provide emotional support. These AIs are now developing at least basic reasoning and planning abilities, which will unlock an entirely new wave of AI development that will radically impact employment and accelerate scientific discovery. These are the “brains” of the machines, and these brains are developing toward human-level+ intelligence. We can download open-course AI “models”/”brains” nearly as powerful as ChatGPT3.5 and Claude2 and run them locally on our computers. You do not need a computer science background to do this. Free brains!

Significant advances in robotics technologies have been made, and factories have opened to produce robots. These are the bodies of the machines. And they are starting to emerge in biologically synthetic form.

Yes, my son will go up into a world of robots. His friends at school are already interacting with them to some degree. I hope we take seriously the importance of helping students prepare for this world.

If we don’t, what are we doing?

---

Additional Microsoft resources on the Future of Work.